PROJECT SUMMARY Metabolomics offers a comprehensive analysis of thousands of small molecules in biological samples. It is a crucial component of systems biology, which characterizes an organism as an integrated and interactive network of various biomolecules. In particular, metabolomic characterization of complex diseases such as hepatocellular carcinoma (HCC) plays an indispensable role in the growing systems biology approaches to identify reliable biomarkers and improved disease treatment strategies. Liquid chromatography coupled to mass spectrometry (LC-MS) and gas chromatography coupled to mass spectrometry (GC-MS) have been extensively used for high- throughput comparison of the levels of thousands of metabolites among biological samples. However, the potential values of many HCC-associated analytes discovered by these platforms have been inadequately explored in systems biology research due to lack of computational tools and resources to: (1) accurately determine the identity of most of the analytes; (2) investigate the rewiring and conserved interactions among metabolites in the progression of the disease, and (3) integrate multi-omic data to evaluate the relationship between disease and metabolites at the systems level. Partly due to these limitations, poor reproducibility of previously identified metabolite biomarker candidates for HCC has been observed, especially when they are evaluated through independent platforms and validation sets. Therefore, new methods are sought to find more potent biomarkers for HCC. This project aims to address this challenge with the help of network-based approaches for: (1) prioritizing putative IDs to assist in metabolite identification; (2) performing differential analysis to uncover relationships between HCC and metabolites; and (3) integrating metabolomic data with transcriptomic, proteomic, and glycomic data to identify highly promising metabolites as biomarker candidates. The clinical relevance of these candidates will be evaluated using isotope dilution by selected reaction monitoring (SRM) and selected ion monitoring (SIM) in sera collected from independent subjects recruited in multiple sites. Also, the performance of the candidates in detecting HCC will be compared against established blood biomarkers such as alpha-fetoprotein (AFP), AFP-L3, des-gamma-carboxy prothrombin (DCP), and golgi protein-73 (GP73). In summary, this project seeks to find metabolite biomarkers for HCC by capitalizing on the power of network modeling, multi-omic data integration, targeted quantitative analysis, and a multicenter repository of biospecimens. We strongly believe that the project will lead to reliable serological biomarkers that are likely to succeed in future large-scale biomarker validation studies.